compiler/rustc_expand/src/mbe/quoted.rs

   1 use crate::mbe::macro_parser::count_metavar_decls;
   2 use crate::mbe::{Delimited, KleeneOp, KleeneToken, MetaVarExpr, SequenceRepetition, TokenTree};
   3
   4 use rustc_ast::token::{self, Delimiter, Token};
   5 use rustc_ast::{tokenstream, NodeId};
   6 use rustc_ast_pretty::pprust;
   7 use rustc_feature::Features;
   8 use rustc_session::parse::{feature_err, ParseSess};
   9 use rustc_span::symbol::{kw, sym, Ident};
  10
  11 use rustc_span::edition::Edition;
  12 use rustc_span::{Span, SyntaxContext};
  13
  14 const VALID_FRAGMENT_NAMES_MSG: &str = "valid fragment specifiers are \
  15                                         `ident`, `block`, `stmt`, `expr`, `pat`, `ty`, `lifetime`, \
  16                                         `literal`, `path`, `meta`, `tt`, `item` and `vis`";
  17
  18 /// Takes a `tokenstream::TokenStream` and returns a `Vec<self::TokenTree>`. Specifically, this
  19 /// takes a generic `TokenStream`, such as is used in the rest of the compiler, and returns a
  20 /// collection of `TokenTree` for use in parsing a macro.
  21 ///
  22 /// # Parameters
  23 ///
  24 /// - `input`: a token stream to read from, the contents of which we are parsing.
  25 /// - `parsing_patterns`: `parse` can be used to parse either the "patterns" or the "body" of a
  26 ///   macro. Both take roughly the same form _except_ that:
  27 ///   - In a pattern, metavars are declared with their "matcher" type. For example `$var:expr` or
  28 ///     `$id:ident`. In this example, `expr` and `ident` are "matchers". They are not present in the
  29 ///     body of a macro rule -- just in the pattern.
  30 ///   - Metavariable expressions are only valid in the "body", not the "pattern".
  31 /// - `sess`: the parsing session. Any errors will be emitted to this session.
  32 /// - `node_id`: the NodeId of the macro we are parsing.
  33 /// - `features`: language features so we can do feature gating.
  34 ///
  35 /// # Returns
  36 ///
  37 /// A collection of `self::TokenTree`. There may also be some errors emitted to `sess`.
  38 pub(super) fn parse(
  39     input: tokenstream::TokenStream,
  40     parsing_patterns: bool,
  41     sess: &ParseSess,
  42     node_id: NodeId,
  43     features: &Features,
  44     edition: Edition,
  45 ) -> Vec<TokenTree> {
  46     // Will contain the final collection of `self::TokenTree`
  47     let mut result = Vec::new();
  48
  49     // For each token tree in `input`, parse the token into a `self::TokenTree`, consuming
  50     // additional trees if need be.
  51     let mut trees = input.into_trees();
  52     while let Some(tree) = trees.next() {
  53         // Given the parsed tree, if there is a metavar and we are expecting matchers, actually
  54         // parse out the matcher (i.e., in `$id:ident` this would parse the `:` and `ident`).
  55         let tree = parse_tree(tree, &mut trees, parsing_patterns, sess, node_id, features, edition);
  56         match tree {
  57             TokenTree::MetaVar(start_sp, ident) if parsing_patterns => {
  58                 let span = match trees.next() {
  59                     Some(tokenstream::TokenTree::Token(Token { kind: token::Colon, span }, _)) => {
  60                         match trees.next() {
  61                             Some(tokenstream::TokenTree::Token(token, _)) => match token.ident() {
  62                                 Some((frag, _)) => {
  63                                     let span = token.span.with_lo(start_sp.lo());
  64
  65                                     let kind =
  66                                         token::NonterminalKind::from_symbol(frag.name, || {
  67                                             // FIXME(#85708) - once we properly decode a foreign
  68                                             // crate's `SyntaxContext::root`, then we can replace
  69                                             // this with just `span.edition()`. A
  70                                             // `SyntaxContext::root()` from the current crate will
  71                                             // have the edition of the current crate, and a
  72                                             // `SyntaxContext::root()` from a foreign crate will
  73                                             // have the edition of that crate (which we manually
  74                                             // retrieve via the `edition` parameter).
  75                                             if span.ctxt() == SyntaxContext::root() {
  76                                                 edition
  77                                             } else {
  78                                                 span.edition()
  79                                             }
  80                                         })
  81                                         .unwrap_or_else(
  82                                             || {
  83                                                 let msg = format!(
  84                                                     "invalid fragment specifier `{}`",
  85                                                     frag.name
  86                                                 );
  87                                                 sess.span_diagnostic
  88                                                     .struct_span_err(span, &msg)
  89                                                     .help(VALID_FRAGMENT_NAMES_MSG)
  90                                                     .emit();
  91                                                 token::NonterminalKind::Ident
  92                                             },
  93                                         );
  94                                     result.push(TokenTree::MetaVarDecl(span, ident, Some(kind)));
  95                                     continue;
  96                                 }
  97                                 _ => token.span,
  98                             },
  99                             tree => tree.as_ref().map_or(span, tokenstream::TokenTree::span),
 100                         }
 101                     }
 102                     tree => tree.as_ref().map_or(start_sp, tokenstream::TokenTree::span),
 103                 };
 104
 105                 result.push(TokenTree::MetaVarDecl(span, ident, None));
 106             }
 107
 108             // Not a metavar or no matchers allowed, so just return the tree
 109             _ => result.push(tree),
 110         }
 111     }
 112     result
 113 }
 114
 115 /// Asks for the `macro_metavar_expr` feature if it is not already declared
 116 fn maybe_emit_macro_metavar_expr_feature(features: &Features, sess: &ParseSess, span: Span) {
 117     if !features.macro_metavar_expr {
 118         let msg = "meta-variable expressions are unstable";
 119         feature_err(&sess, sym::macro_metavar_expr, span, msg).emit();
 120     }
 121 }
 122
 123 /// Takes a `tokenstream::TokenTree` and returns a `self::TokenTree`. Specifically, this takes a
 124 /// generic `TokenTree`, such as is used in the rest of the compiler, and returns a `TokenTree`
 125 /// for use in parsing a macro.
 126 ///
 127 /// Converting the given tree may involve reading more tokens.
 128 ///
 129 /// # Parameters
 130 ///
 131 /// - `tree`: the tree we wish to convert.
 132 /// - `outer_trees`: an iterator over trees. We may need to read more tokens from it in order to finish
 133 ///   converting `tree`
 134 /// - `parsing_patterns`: same as [parse].
 135 /// - `sess`: the parsing session. Any errors will be emitted to this session.
 136 /// - `features`: language features so we can do feature gating.
 137 fn parse_tree(
 138     tree: tokenstream::TokenTree,
 139     outer_trees: &mut impl Iterator<Item = tokenstream::TokenTree>,
 140     parsing_patterns: bool,
 141     sess: &ParseSess,
 142     node_id: NodeId,
 143     features: &Features,
 144     edition: Edition,
 145 ) -> TokenTree {
 146     // Depending on what `tree` is, we could be parsing different parts of a macro
 147     match tree {
 148         // `tree` is a `$` token. Look at the next token in `trees`
 149         tokenstream::TokenTree::Token(Token { kind: token::Dollar, span }, _) => {
 150             // FIXME: Handle `Invisible`-delimited groups in a more systematic way
 151             // during parsing.
 152             let mut next = outer_trees.next();
 153             let mut trees: Box<dyn Iterator<Item = tokenstream::TokenTree>>;
 154             if let Some(tokenstream::TokenTree::Delimited(_, Delimiter::Invisible, tts)) = next {
 155                 trees = Box::new(tts.into_trees());
 156                 next = trees.next();
 157             } else {
 158                 trees = Box::new(outer_trees);
 159             }
 160
 161             match next {
 162                 // `tree` is followed by a delimited set of token trees.
 163                 Some(tokenstream::TokenTree::Delimited(delim_span, delim, tts)) => {
 164                     if parsing_patterns {
 165                         if delim != Delimiter::Parenthesis {
 166                             span_dollar_dollar_or_metavar_in_the_lhs_err(
 167                                 sess,
 168                                 &Token { kind: token::OpenDelim(delim), span: delim_span.entire() },
 169                             );
 170                         }
 171                     } else {
 172                         match delim {
 173                             Delimiter::Brace => {
 174                                 // The delimiter is `{`.  This indicates the beginning
 175                                 // of a meta-variable expression (e.g. `${count(ident)}`).
 176                                 // Try to parse the meta-variable expression.
 177                                 match MetaVarExpr::parse(&tts, delim_span.entire(), sess) {
 178                                     Err(mut err) => {
 179                                         err.emit();
 180                                         // Returns early the same read `$` to avoid spanning
 181                                         // unrelated diagnostics that could be performed afterwards
 182                                         return TokenTree::token(token::Dollar, span);
 183                                     }
 184                                     Ok(elem) => {
 185                                         maybe_emit_macro_metavar_expr_feature(
 186                                             features,
 187                                             sess,
 188                                             delim_span.entire(),
 189                                         );
 190                                         return TokenTree::MetaVarExpr(delim_span, elem);
 191                                     }
 192                                 }
 193                             }
 194                             Delimiter::Parenthesis => {}
 195                             _ => {
 196                                 let tok = pprust::token_kind_to_string(&token::OpenDelim(delim));
 197                                 let msg = format!("expected `(` or `{{`, found `{}`", tok);
 198                                 sess.span_diagnostic.span_err(delim_span.entire(), &msg);
 199                             }
 200                         }
 201                     }
 202                     // If we didn't find a metavar expression above, then we must have a
 203                     // repetition sequence in the macro (e.g. `$(pat)*`).  Parse the
 204                     // contents of the sequence itself
 205                     let sequence = parse(tts, parsing_patterns, sess, node_id, features, edition);
 206                     // Get the Kleene operator and optional separator
 207                     let (separator, kleene) =
 208                         parse_sep_and_kleene_op(&mut trees, delim_span.entire(), sess);
 209                     // Count the number of captured "names" (i.e., named metavars)
 210                     let num_captures =
 211                         if parsing_patterns { count_metavar_decls(&sequence) } else { 0 };
 212                     TokenTree::Sequence(
 213                         delim_span,
 214                         SequenceRepetition { tts: sequence, separator, kleene, num_captures },
 215                     )
 216                 }
 217
 218                 // `tree` is followed by an `ident`. This could be `$meta_var` or the `$crate`
 219                 // special metavariable that names the crate of the invocation.
 220                 Some(tokenstream::TokenTree::Token(token, _)) if token.is_ident() => {
 221                     let (ident, is_raw) = token.ident().unwrap();
 222                     let span = ident.span.with_lo(span.lo());
 223                     if ident.name == kw::Crate && !is_raw {
 224                         TokenTree::token(token::Ident(kw::DollarCrate, is_raw), span)
 225                     } else {
 226                         TokenTree::MetaVar(span, ident)
 227                     }
 228                 }
 229
 230                 // `tree` is followed by another `$`. This is an escaped `$`.
 231                 Some(tokenstream::TokenTree::Token(Token { kind: token::Dollar, span }, _)) => {
 232                     if parsing_patterns {
 233                         span_dollar_dollar_or_metavar_in_the_lhs_err(
 234                             sess,
 235                             &Token { kind: token::Dollar, span },
 236                         );
 237                     } else {
 238                         maybe_emit_macro_metavar_expr_feature(features, sess, span);
 239                     }
 240                     TokenTree::token(token::Dollar, span)
 241                 }
 242
 243                 // `tree` is followed by some other token. This is an error.
 244                 Some(tokenstream::TokenTree::Token(token, _)) => {
 245                     let msg = format!(
 246                         "expected identifier, found `{}`",
 247                         pprust::token_to_string(&token),
 248                     );
 249                     sess.span_diagnostic.span_err(token.span, &msg);
 250                     TokenTree::MetaVar(token.span, Ident::empty())
 251                 }
 252
 253                 // There are no more tokens. Just return the `$` we already have.
 254                 None => TokenTree::token(token::Dollar, span),
 255             }
 256         }
 257
 258         // `tree` is an arbitrary token. Keep it.
 259         tokenstream::TokenTree::Token(token, _) => TokenTree::Token(token),
 260
 261         // `tree` is the beginning of a delimited set of tokens (e.g., `(` or `{`). We need to
 262         // descend into the delimited set and further parse it.
 263         tokenstream::TokenTree::Delimited(span, delim, tts) => TokenTree::Delimited(
 264             span,
 265             Delimited {
 266                 delim,
 267                 tts: parse(tts, parsing_patterns, sess, node_id, features, edition),
 268             },
 269         ),
 270     }
 271 }
 272
 273 /// Takes a token and returns `Some(KleeneOp)` if the token is `+` `*` or `?`. Otherwise, return
 274 /// `None`.
 275 fn kleene_op(token: &Token) -> Option<KleeneOp> {
 276     match token.kind {
 277         token::BinOp(token::Star) => Some(KleeneOp::ZeroOrMore),
 278         token::BinOp(token::Plus) => Some(KleeneOp::OneOrMore),
 279         token::Question => Some(KleeneOp::ZeroOrOne),
 280         _ => None,
 281     }
 282 }
 283
 284 /// Parse the next token tree of the input looking for a KleeneOp. Returns
 285 ///
 286 /// - Ok(Ok((op, span))) if the next token tree is a KleeneOp
 287 /// - Ok(Err(tok, span)) if the next token tree is a token but not a KleeneOp
 288 /// - Err(span) if the next token tree is not a token
 289 fn parse_kleene_op(
 290     input: &mut impl Iterator<Item = tokenstream::TokenTree>,
 291     span: Span,
 292 ) -> Result<Result<(KleeneOp, Span), Token>, Span> {
 293     match input.next() {
 294         Some(tokenstream::TokenTree::Token(token, _)) => match kleene_op(&token) {
 295             Some(op) => Ok(Ok((op, token.span))),
 296             None => Ok(Err(token)),
 297         },
 298         tree => Err(tree.as_ref().map_or(span, tokenstream::TokenTree::span)),
 299     }
 300 }
 301
 302 /// Attempt to parse a single Kleene star, possibly with a separator.
 303 ///
 304 /// For example, in a pattern such as `$(a),*`, `a` is the pattern to be repeated, `,` is the
 305 /// separator, and `*` is the Kleene operator. This function is specifically concerned with parsing
 306 /// the last two tokens of such a pattern: namely, the optional separator and the Kleene operator
 307 /// itself. Note that here we are parsing the _macro_ itself, rather than trying to match some
 308 /// stream of tokens in an invocation of a macro.
 309 ///
 310 /// This function will take some input iterator `input` corresponding to `span` and a parsing
 311 /// session `sess`. If the next one (or possibly two) tokens in `input` correspond to a Kleene
 312 /// operator and separator, then a tuple with `(separator, KleeneOp)` is returned. Otherwise, an
 313 /// error with the appropriate span is emitted to `sess` and a dummy value is returned.
 314 fn parse_sep_and_kleene_op(
 315     input: &mut impl Iterator<Item = tokenstream::TokenTree>,
 316     span: Span,
 317     sess: &ParseSess,
 318 ) -> (Option<Token>, KleeneToken) {
 319     // We basically look at two token trees here, denoted as #1 and #2 below
 320     let span = match parse_kleene_op(input, span) {
 321         // #1 is a `?`, `+`, or `*` KleeneOp
 322         Ok(Ok((op, span))) => return (None, KleeneToken::new(op, span)),
 323
 324         // #1 is a separator followed by #2, a KleeneOp
 325         Ok(Err(token)) => match parse_kleene_op(input, token.span) {
 326             // #2 is the `?` Kleene op, which does not take a separator (error)
 327             Ok(Ok((KleeneOp::ZeroOrOne, span))) => {
 328                 // Error!
 329                 sess.span_diagnostic.span_err(
 330                     token.span,
 331                     "the `?` macro repetition operator does not take a separator",
 332                 );
 333
 334                 // Return a dummy
 335                 return (None, KleeneToken::new(KleeneOp::ZeroOrMore, span));
 336             }
 337
 338             // #2 is a KleeneOp :D
 339             Ok(Ok((op, span))) => return (Some(token), KleeneToken::new(op, span)),
 340
 341             // #2 is a random token or not a token at all :(
 342             Ok(Err(Token { span, .. })) | Err(span) => span,
 343         },
 344
 345         // #1 is not a token
 346         Err(span) => span,
 347     };
 348
 349     // If we ever get to this point, we have experienced an "unexpected token" error
 350     sess.span_diagnostic.span_err(span, "expected one of: `*`, `+`, or `?`");
 351
 352     // Return a dummy
 353     (None, KleeneToken::new(KleeneOp::ZeroOrMore, span))
 354 }
 355
 356 // `$$` or a meta-variable is the lhs of a macro but shouldn't.
 357 //
 358 // For example, `macro_rules! foo { ( ${length()} ) => {} }`
 359 fn span_dollar_dollar_or_metavar_in_the_lhs_err<'sess>(sess: &'sess ParseSess, token: &Token) {
 360     sess.span_diagnostic
 361         .span_err(token.span, &format!("unexpected token: {}", pprust::token_to_string(token)));
 362     sess.span_diagnostic.span_note_without_error(
 363         token.span,
 364         "`$$` and meta-variable expressions are not allowed inside macro parameter definitions",
 365     );
 366 }